The superionic transition in lithium oxide $\left({\mathrm{Li}}_2\mathrm{O}\right)$ presents significant challenges in establishing the structure-property relationship between its microscopic dynamical behavior to macroscopic physical properties. Additionally, the fundamental mechanism driving this transition remains contentious due to competing theoretical interpretations. Using machine learning potentials that maintain ab initio-level precision while enabling extended in both temporal and spatial scales, we systematically investigate lithium-ion diffusion dynamics spanning from crystalline to superionic phases. Our results give a clear physical pattern of ${\mathrm{Li}}_2\mathrm{O}$ Arrhenius plot throughout three distinct regions. The elastic softening serves as a signal of accelerating of Li-ions movement. Superionic transition is manifested in the melting of the Li sublattice. Dynamical analysis indicates that the superionic transition is driven by the formation of Frenkel pairs. Octahedral interstitial Li-ions, increasing sharply near the superionic state, act as primary diffusion carriers. The tetrahedral Li-ions exhibit much lower individual mobility and play a secondary role in diffusion under superionic state. This highlights the unique defect-mediated nature of diffusion in superionic state.

Close

PrevNext

View figure in article

• Diffusion
• Ionic transport
• Energy materials
• Density functional calculations
• Machine learning
• Molecular dynamics